Deep neural networks (DNNs) have become a critical tool in natural language processing (NLP) for a wide variety of language technologies, from syntax to semantics to pragmatics. In particular, in the field of natural language inference (NLI), DNNs have become the de-facto model, providing significantly better results than previous paradigms. Their power lies in their ability to embed complex language ambiguities in high dimensional spaces coupled with non-linear compositional transformations learned to directly optimize task-specific objective functions. We propose to adapt Deep NLI techniques to the biomedical domain, specifically investigating question answering, information extraction and synthesis. The biomedical domain presents many key challenges and a critical impact that standard NLI challenges do not posses. First, while standard NLI data sets requires a system to model basic world knowledge (e.g., that ‘soccer’ is a ‘sport’), they do not presume a rich domain knowledge encoded in various and often heterogeneous resources such as scientific articles, textbooks and structured databases. Second, while standard NLI data sets presume that the answer/inference is encoded in a single utterance, the ability to reason and extract information from biomedical domains often requires information synthesis from multiple utterances, paragraphs, and even documents. Finally, whereas standard NLI is a broad challenge aimed at testing whether computers can make general inferences in language, biomedical texts are a grounded and impactful domain where progress in automated reasoning will directly impact the efficacy of researchers, physicians, publishers and policy makers.

People increasingly depend on their personal smart devices (smartphones, smart watches, voice assistants) to communicate, access information, transact, and perform critical functions. These devices overwhelmingly comprise proprietary software and hardware components made by a handful of non-EU vendors. These vendors have exceptional control over those devices, raising justified concerns over anti-competitive practices, massive data gathering, surveillance, denial of service, and consequently personal freedoms and democracy. Digital sovereignty has been identified as a key political priority for the EU. SPUCS: “software architectures for Secure, Private, and User-Controlled Smart devices” contributes novel privacy-preserving architectures, realizable on existing hardware, that limit the control of vendors on user devices towards a future of open, sovereign personal devices. While previous initiatives for open and privacy-preserving smart devices lacked in functionality and therefore in adoption, SPUCS introduces concepts and methods that enable users to continue using their preferred operating systems and applications in an isolated way, with increased privacy protection. Specifically, the project will contribute secure mechanisms to seamlessly run multiple mobile operating systems simultaneously on a mobile device with strong isolation and privacy guarantees – limiting vendors’ ability to collect and connect data. Furthermore, the project will develop methods for securely running individual applications (phone, messaging, banking) on an open platform without trusting the operating system or proprietary vendor software. Finally, the project will plan an adoption strategy that will boost the scientific and industrial impact of the results towards a secure and self-sovereign Europe. The project also fosters interdisciplinary two-way transfer of knowledge between the researcher and the host, as well as training of the researcher in advanced scientific techniques.

The ultimate goal of this Fellowship, titled “Bayesian infEReNce And moDel sElecTion for sTochastic Epidemics” (BERNADETTE), is to train a talented researcher through a research project focused on the development of novel statistical methodology for the modeling of infectious diseases like COVID-19. The success of the interdisciplinary project will lead to a number of multidisciplinary innovations in epidemiology, Public Health policy and statistics, which will contribute to the timely identification of optimal disease control strategies. The Fellow – Dr. Lampros Bouranis – will be trained in the fields of statistics and epidemiology, receiving access to a unique training experience at the host – Department of Statistics, Athens University of Economics and Business (AUEB) – and co-hosts. The BERNADETTE outputs will be relevant to healthcare and the EU Epidemic intelligence, by: i) offering novel statistical methodology for the analysis of COVID-19 outbreak data and the description of a number of aspects of the underlying infection pathway of the disease, ii) quantifying the effect of non-pharmaceutical interventions based on an epidemic model, iii) allowing for the forecasting of future case number scenarios, iv) contributing in the assessment of the socio-economic impact of different response strategies for human epidemics in Europe in order to improve European preparedness planning and support decision-making in the framework of national epidemic preparedness plans. The BERNADETTE outputs will contribute to the enhancement of EU scientific excellence. Additionally, the project will enable the establishment of a long-term collaboration between the host and co-hosts, bringing the centers of European research excellence together.